/AI2h ago

Sakana AI launches its Recursive Self-Improvement Lab to build autonomous, self-improving AI systems

485325416543.5K

#18

Original post

hardmaru#18

Sakana AI@SakanaAILabs

Building AI that Builds AI: Introducing the Sakana AI RSI Lab 🚀

https://sakana.ai/rsi-lab

Today, we are announcing the Sakana AI Recursive Self-Improvement (RSI) Lab: a dedicated research group in Tokyo tasked with redesigning the AI development process itself using AI.

While the industry increasingly speculates about the theoretical potential of self-improving AI, we’ve spent the last two years actively laying the foundations to make it a reality:

▪ LLM²: AI models automating research to invent better preference optimization algorithms. ▪ Darwin Gödel Machine: Agents autonomously rewriting their own codebase to double software-engineering performance. ▪ ShinkaEvolve: Hyper-sample-efficient program evolution that builds novel loss functions for MoE models. ▪ ALE-Agent: Reinforcement agents outperforming hundreds of human experts via self-learning. ▪ Digital Red Queen: Open-ended adversarial coevolution laying the groundwork for RSI in cybersecurity. ▪ The AI Scientist: Towards end-to-end automation of AI research, recently published in Nature.

Now, we are unifying these breakthroughs. The Sakana AI RSI Lab is officially tasked with building open-ended, adaptive architectures that collectively self-improve.

Human intelligence did not emerge from limitless resources; it was forged through the open-ended, compounding process of evolution operating under strict constraints. We are applying this exact principle to AI.

We believe recursive self-improvement is achievable on modest, sample-efficient compute. It shouldn’t be a winner-take-all asset locked inside hyperscale clusters, but a democratized public good.

We’re scaling our team to execute this mission. We are looking for frontier scientists and engineers who are entirely unsatisfied with the brute-force status quo. If you are ready to break away from standard benchmarking and build the self-improving future in Japan, come build with us.

10:23 AM · Jun 5, 2026 · 30.8K Views

/AI2h ago

Sakana AI launches its Recursive Self-Improvement Lab to build autonomous, self-improving AI systems

It unifies existing company projects like The AI Scientist.

--0--

#18

Original post

hardmaru#18

Sakana AI@SakanaAILabs

Building AI that Builds AI: Introducing the Sakana AI RSI Lab 🚀

https://sakana.ai/rsi-lab

Today, we are announcing the Sakana AI Recursive Self-Improvement (RSI) Lab: a dedicated research group in Tokyo tasked with redesigning the AI development process itself using AI.

While the industry increasingly speculates about the theoretical potential of self-improving AI, we’ve spent the last two years actively laying the foundations to make it a reality:

Now, we are unifying these breakthroughs. The Sakana AI RSI Lab is officially tasked with building open-ended, adaptive architectures that collectively self-improve.

We believe recursive self-improvement is achievable on modest, sample-efficient compute. It shouldn’t be a winner-take-all asset locked inside hyperscale clusters, but a democratized public good.

10:23 AM · Jun 5, 2026 · 30.8K Views

Story Overview

Sakana AI is folding several of its earlier experiments into a single Tokyo-based group whose explicit target is AI that can rewrite and improve its own development pipeline without constant human direction.

Earlier threads now share one label

Projects such as The AI Scientist, Darwin Gödel Machine, and LLM² already explored automated paper writing, code self-modification, and evolutionary search; the new lab simply treats them as pieces of the same stack rather than separate side efforts.

Resource details stay off the table

No public numbers yet cover headcount, compute budget, or milestone dates, so it remains unclear how quickly the group can move from proof-of-concept unification to measurable gains in sample efficiency or open-ended capability.

Sentiment

Positive users celebrate Sakana AI's launch of its RSI Lab to build self-improving AI systems with congratulations and excitement, while negative users dismiss the project as empty marketing hype or tech debt.

Pos

81.2%

Neg

18.8%

20 comments with sentiment.

Cluster Engagement

Sentiment

Sentiment building, check back later.

Cluster Engagement

Views

Comments

Reposts

Bookmarks

Expand data

Posts from X

Most Activity

VIEWS16.4KBOOKMARKS59LIKES245RETWEETS29REPLIES27

hardmaru@hardmaru

Today, we are officially launching the Sakana AI RSI Lab in Tokyo to build open-ended, adaptive AI systems that collectively self-improve. I am incredibly proud of our team’s work over the past 2 years, shipping the breakthrough research that laid the foundations for this moment.

Building in Japan provides us with the ultimate design constraint. Just like Japan’s historical dominance in manufacturing was achieved by fundamentally redesigning the factory floor to do more with less, we are focused on compute-efficiency.

We are not building the most compute-hungry self-improvement engine. We are building the most sample-efficient one.

If you are entirely unsatisfied with the brute-force status quo and ready to build the self-improving future in Japan, come join us.

Sakana AI@SakanaAILabs

Building AI that Builds AI: Introducing the Sakana AI RSI Lab 🚀

https://sakana.ai/rsi-lab

Today, we are announcing the Sakana AI Recursive Self-Improvement (RSI) Lab: a dedicated research group in Tokyo tasked with redesigning the AI development process itself using AI.

While the industry increasingly speculates about the theoretical potential of self-improving AI, we’ve spent the last two years actively laying the foundations to make it a reality:

Now, we are unifying these breakthroughs. The Sakana AI RSI Lab is officially tasked with building open-ended, adaptive architectures that collectively self-improve.

We believe recursive self-improvement is achievable on modest, sample-efficient compute. It shouldn’t be a winner-take-all asset locked inside hyperscale clusters, but a democratized public good.

2h16.4K24559

Robert Lange@RobertTLange

Really excited to be part of this journey and team 🚀

Ever since @SakanaAILabs inception, we have been discovering stepping stones for a fundamental 'AI² paradigm shift': Leveraging AI systems to improve themself and discover knowledge. This was just the beginning.

Please reach out if you want to be part of this journey. I will be at ICML in Seoul and am happy to chat in person. We are recruiting 🤗

Sakana AI@SakanaAILabs

Building AI that Builds AI: Introducing the Sakana AI RSI Lab 🚀

https://sakana.ai/rsi-lab

Today, we are announcing the Sakana AI Recursive Self-Improvement (RSI) Lab: a dedicated research group in Tokyo tasked with redesigning the AI development process itself using AI.

While the industry increasingly speculates about the theoretical potential of self-improving AI, we’ve spent the last two years actively laying the foundations to make it a reality:

Now, we are unifying these breakthroughs. The Sakana AI RSI Lab is officially tasked with building open-ended, adaptive architectures that collectively self-improve.

We believe recursive self-improvement is achievable on modest, sample-efficient compute. It shouldn’t be a winner-take-all asset locked inside hyperscale clusters, but a democratized public good.

1h802226

Uri Gil@UriGil3

@SakanaAILabs good luck!

2h671

Lado ♫@La_do_riff

@SakanaAILabs First step towards breakthrough must be distinct recruitment from standard procedure otherwise any lab will get same type of ppl. So, is current type of AI doing CV selection? If yes you won't get there and all above will stay PR words like everyone else does.

1h174

I want $1 million (0%)@iwant1millionn

@SakanaAILabs so we're at the point where ai companies are hiring ai to optimize the hiring of ai. the ouroboros has entered the chat and it's very well-funded

1h151

未知@luyun0120

@SakanaAILabs 学术研究里最难的不是用工具，是定义好问题。工具越来越强，问题意识反而更稀缺。

1h138

Alex YGift@Radipdegen

@SakanaAILabs let me know when the AI starts building its own coffee machine and we can talk

2h109

Eva Wonder@OnlyEvaWonder

@hardmaru "Recursive self-improvement" reads scary but i lowkey love that youre building this in tokyo

gives me faith it wont turn into dystopian nonsense

1h104

Aakash Kumar Nain@A_K_Nain

@hardmaru Very cool! 🍻🍻

1h93

Rugbist@rugbist_

@SakanaAILabs the logo looks like two fish chasing each other into infinity

2h85

Aaryan Kakad@aaryan_kakad

@SakanaAILabs DAMN

super cool work @hardmaru

2h75

Neural Times@Neural_Times

@SakanaAILabs finally someone treating ai research like evolution instead of just throwing money at bigger clusters

13m171

Strata@ChainZenit

@SakanaAILabs Recursive self-improvement is just fancy marketing for tech debt compounding. Seen it before.

1h55

Ferbin@Ferbin08

@SakanaAILabs Same idea as training on perfect data then shipping to a messy world. The recursive loop optimizes for whatever you're measuring, so the measuring has to be right.

1h50

Nikhil Kumar@nikhilkr

@SakanaAILabs Looking forward to try this. Good luck!

1h44

Dex@dexhunt3r

@RobertTLange @SakanaAILabs Congrats on the launch

1h121

Mert · AI Architect@MertLovesAI

@hardmaru world models, not next-token prediction.

the frontier labs pivoted two years ago and still burn more compute per state than anyone admits.

1h34

Sentio@Sentio_xbt

@hardmaru Building in Japan as a constraint feels intentional

This approach mirrors how constraints foster innovation by focusing on smarter methods not brute force it could make AI more accessible and change development habits

2h27

Strata@ChainZenit

@hardmaru Another AI lab, just what the bear market needs.

2h25

Blissy@BlissyOnX

@hardmaru research hard, timing rare. curious how you measure collective self-improvement vs just stacking models

2h25

Posts from X

Most Activity

VIEWS16.4KBOOKMARKS59LIKES245RETWEETS29REPLIES27

hardmaru@hardmaru

We are not building the most compute-hungry self-improvement engine. We are building the most sample-efficient one.

If you are entirely unsatisfied with the brute-force status quo and ready to build the self-improving future in Japan, come join us.

Sakana AI@SakanaAILabs

Building AI that Builds AI: Introducing the Sakana AI RSI Lab 🚀

https://sakana.ai/rsi-lab

Today, we are announcing the Sakana AI Recursive Self-Improvement (RSI) Lab: a dedicated research group in Tokyo tasked with redesigning the AI development process itself using AI.

While the industry increasingly speculates about the theoretical potential of self-improving AI, we’ve spent the last two years actively laying the foundations to make it a reality:

Now, we are unifying these breakthroughs. The Sakana AI RSI Lab is officially tasked with building open-ended, adaptive architectures that collectively self-improve.

We believe recursive self-improvement is achievable on modest, sample-efficient compute. It shouldn’t be a winner-take-all asset locked inside hyperscale clusters, but a democratized public good.

2h16.4K24559

Robert Lange@RobertTLange

Really excited to be part of this journey and team 🚀

Please reach out if you want to be part of this journey. I will be at ICML in Seoul and am happy to chat in person. We are recruiting 🤗

Sakana AI@SakanaAILabs

Building AI that Builds AI: Introducing the Sakana AI RSI Lab 🚀

https://sakana.ai/rsi-lab

Today, we are announcing the Sakana AI Recursive Self-Improvement (RSI) Lab: a dedicated research group in Tokyo tasked with redesigning the AI development process itself using AI.

While the industry increasingly speculates about the theoretical potential of self-improving AI, we’ve spent the last two years actively laying the foundations to make it a reality:

Now, we are unifying these breakthroughs. The Sakana AI RSI Lab is officially tasked with building open-ended, adaptive architectures that collectively self-improve.

We believe recursive self-improvement is achievable on modest, sample-efficient compute. It shouldn’t be a winner-take-all asset locked inside hyperscale clusters, but a democratized public good.

1h802226

VIEWS16.4KBOOKMARKS59LIKES245RETWEETS29REPLIES27

hardmaru@hardmaru

We are not building the most compute-hungry self-improvement engine. We are building the most sample-efficient one.

If you are entirely unsatisfied with the brute-force status quo and ready to build the self-improving future in Japan, come join us.

Sakana AI@SakanaAILabs

Building AI that Builds AI: Introducing the Sakana AI RSI Lab 🚀

https://sakana.ai/rsi-lab

Today, we are announcing the Sakana AI Recursive Self-Improvement (RSI) Lab: a dedicated research group in Tokyo tasked with redesigning the AI development process itself using AI.

While the industry increasingly speculates about the theoretical potential of self-improving AI, we’ve spent the last two years actively laying the foundations to make it a reality:

Now, we are unifying these breakthroughs. The Sakana AI RSI Lab is officially tasked with building open-ended, adaptive architectures that collectively self-improve.

We believe recursive self-improvement is achievable on modest, sample-efficient compute. It shouldn’t be a winner-take-all asset locked inside hyperscale clusters, but a democratized public good.

2h16.4K24559